Eff ects of Climate and Competition for Off shore Prey on Growth, Survival, and Reproductive Potential of Coho Salmon in Southeast Alaska - Alaska ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
North Pacific Anadromous Fish Commission Bulletin No. 6: 329–347, 2016 Effects of Climate and Competition for Offshore Prey on Growth, Survival, and Reproductive Potential of Coho Salmon in Southeast Alaska Leon D. Shaul1 and Harold J. Geiger2 Alaska Department of Fish and Game, Division of Commercial Fisheries, 1 P.O. Box 110024, Douglas, AK 99811-0024, USA 2 St. Hubert Research Group, 222 Seward Street, Suite 205, Juneau, AK 99801, USA Shaul, L.D., and H.J. Geiger. 2016. Effects of climate and competition for offshore prey on growth, survival, and reproductive potential of coho salmon in Southeast Alaska. N. Pac. Anadr. Fish Comm. Bull. 6: 329–347. doi:10.23849/npafcb6/329.347. Abstract: In the offshore Gulf of Alaska (GOA), coho salmon exhibit strong dependence upon a single prey species, the minimal armhook squid (Berryteuthis anonychus). We propose and then test elements of the general hypothesis that coho salmon adult size in Southeast Alaska reflects predator-prey interactions among coho salmon, pink salmon, and squid, where squid are the main prey of coho salmon while pink salmon mediate squid abundance as both competitors and predators of squid. The majority (65%) of variation in size of coho salmon over a 45-year period was explained equally by the catch biomass of pink salmon in the GOA and by the PDO index during squid emergence and development, averaged at lags in 2-year increments (matching the life cycles of pink salmon and squid) of up to four years. We extend the analysis to examine effects on marine survival, sex ratio, and per capita reproductive potential and examine evidence for growth-related late-marine mortality. Our results lend support for an important late-marine period for coho salmon survival and for the role of pink salmon as a keystone predator that controls the trophic structure of salmon forage and the flow of energy in the offshore GOA ecosystem. Our findings also indicate that the capacity of the GOA to produce pink salmon for harvest, while maintaining stable adult coho salmon weight (based on inferred stable squid prey populations), is highly variable and closely linked with atmospheric forcing. Keywords: coho salmon, Berryteuthis anonychus, squid, pink salmon, growth, survival, climate, competition INTRODUCTION determining their growth at sea, where most spend approx- imately 16 months. Southeast Alaska coho salmon are lim- The relationship among salmon species (Oncorhynchus ited to the northeast Pacific (Myers et al. 1996) where they spp.) and their prey in the offshore Gulf of Alaska (GOA) are dependent upon a single calorie-rich prey species to fuel has been described as a “trophic triangle” in which flexible an exceptionally rapid growth rate during their second sea- planktivores (pink O. gorbusha and sockeye O. nerka salm- son at sea (Ishida et al. 1998). Berryteuthis anonychus has on) function as intra-guild predators that both prey upon min- been shown to be the primary offshore prey of maturing coho imal armhook squid (Berryteuthis anonychus) and compete salmon across varying climate regimes (LeBrasseur 1966; with them for zooplankton prey (Aydin 2000; Uchikawa et al. Pearcy et al. 1988; Davis 2003; Kaeriyama et al. 2004). Da- 2004; Fig. 1). Berryteuthis anonychus is also the predominant vis (2003) found that coho salmon in subarctic waters in the prey of obligate nektivores (coho O. kisutch and Chinook O. central North Pacific consumed almost exclusively large sub- tshawytscha salmon, and steelhead O. mykiss) that feed pri- adult and adult B. anonychus, which comprised the majority marily on squid and (to a lesser extent) fish in these same wa- of the diet of all size classes larger than 500 g, and was highly ters (Kaeriyama et al. 2004; Atcheson et al. 2012). Here, we correlated with stomach fullness. Coho salmon feeding in examine this relationship through size and survival of coho summertime increased their stomach contents index (SCI) salmon in Southeast Alaska. with increasing size, as larger fish were able to catch larger Coho salmon exhibit features that, compared with other squid, thereby further increasing their capacity for growth. salmon species, reduce the range of plausible mechanisms While squid comprised 83% of the prey weight consumed by All correspondence should be addressed to L. Shaul. © 2016 North Pacific Anadromous Fish Commission e-mail: leon.shaul@alaska.gov 329
NPAFC Bulletin No. 6 Shaul and Geiger year weights have remained more stable, increasing from a 1970s average of 3.07 kg to a peak in 1984–1988 (average 3.55 kg) followed by a stable trend (average 3.20 kg) during 1990–2010, before dropping abruptly to 2.69–2.93 kg in 2012–2014. Climatic variability may also be important for growth of coho salmon, either through temperature mediated effects on growth or food web effects on prey (Aydin et al. 2005; Beauchamp 2009). Studies of covariation between coho salmon length and ocean environmental variables have gen- erally found poor correlation in Alaska populations at time Fig. 1. Primary trophic connections between zooplankton and six lags considered to be most important (Hobday and Boehlert species of maturing salmon in offshore waters of the Gulf of Alaska 2001; Wells et al. 2006). However, Wells et al. (2008) ob- (modified from Aydin 2000). served a direct positive relationship between growth and the Aleutian Low Pressure Index (ALPI) in a Southeast Alaska maturing coho salmon, their contribution to digestible calo- population of Chinook salmon, a species with an offshore ries was even greater (93%), after accounting for their high diet comprised primarily of squid (similar to coho salmon; caloric density and digestibility (Davis et al. 1998). Kaeriyama et al. 2004). Intensification of the Aleutian Low, Although its rapid early growth rate and small size at and associated positive phase in the related Pacific Decadal maturity have led most investigators to conclude that B. Oscillation (PDO) index, has been shown to be potential- anonychus has a 1-year lifespan (Nesis 1997; Katugin et al. ly important to growth and survival in early stages in GOA 2005; Drobny et al. 2008), Jorgensen (2011) presents com- fish populations, through increased phytoplankton and zoo- pelling evidence for a 2-year lifespan based upon a consis- plankton production (Brodeur and Ware 1992), potentially tent biennial cycle (over a 19-year period) in abundance of as a result of shallowing of the mixed layer (Polovina et al. paralarvae in the northwestern GOA that was correlated with 1995). In addition, the same climatic pattern is also thought abundance of pink salmon. Pink salmon, which also have to have an important positive effect on transport by currents a 2-year lifespan, have increased in abundance in odd years and subsequent survival of larvae of some marine fish spe- while even-year returns have remained more stable (Fig. 2A). cies (Bailey and Picquelle 2002). Although less studied, at- Average weight of troll-caught coho salmon in South- mospheric forcing may similarly affect growth and survival east Alaska shifted from odd-year to even-year dominance of cephalopod larvae. in 1982–1983 (Fig. 2B), two cycles after an opposite shift Review of literature on offshore salmon feeding ecology in cyclic dominance in the commercial catch of pink salmon and climatic effects on salmon growth led us to hypothesize populations in the GOA (Fig. 2A). Coho salmon averaged that the observed history of average adult weight (Fig. 2B) 5.4% larger in odd years during the first decade of the 45- was influenced by availability of maturing squid, which we year series (1970–1979) but 14.1% smaller during 2005– hypothesized was influenced by bottom-up climate-related 2014. While average weight in odd years declined from a processes controlling squid recruitment and by direct compe- peak of 3.64 kg in 1977 to 2.45–2.60 kg in 2011–2013, even- tition for squid by pink salmon. However, initial exploration $ 3LQN 6DOPRQ&DWFK B%(Coho Salmon Weight) &RKR6DOPRQ :HLJKW 0LOOLRQVRI)LVK :HLJKW NJ (YHQ
Climate and competition effects on coho salmon NPAFC Bulletin No. 6 3,500 months of marine residence (e.g. Holtby et al. 1990; Pearcy 1992; Beamish et al. 2004). However, evidence of such a 3,000 Coho period has remained elusive in studies of growth and sur- vival of coho salmon in Southeast Alaska, where indirect 2,500 Pink evidence has instead favored an important late period for growth and survival after juveniles leave coastal waters late in their first summer at sea. Hobday and Boehlert (2001) Weight (g) 2,000 found that environmental conditions when adults were re- turning explained more variance in survival of Alaska pop- 1,500 ulations compared with the first season at sea. In northern Southeast Alaska, LaCroix et al. (2009) found no relation- 1,000 ship between indices of juvenile coho salmon size, condi- Squid tion, abundance, or biophysical variables and subsequent Increasing marine survival and harvest. Although marine survival of 500 in Diet adult pink salmon and age-.0 jack coho salmon from Auke 0 Creek was correlated with Southeast Alaska coastal ocean response metrics, adult coho salmon marine survival was November December January April March May August February July September June not, suggesting that different factors likely influence surviv- al of adults beyond their seaward migration phase (Orsi et al. 2013). The biennial cycle in size of adult coho salmon Month was not evident in juveniles on 24 July, after approximately Fig. 3. Monthly average weight of coho and pink salmon during 2 months at sea (LaCroix et al. 2009), indicating that the their final months at sea (Ishida et al. 1998) and the approximate difference in apparent growth likely occurs in offshore wa- threshold weight (1,000 g) for pink salmon to begin preying on maturing Berryteuthis anonychus (Aydin 2000; Davis 2003). ters. Scale growth of Auke Creek adults also indicates that size-at-maturity is determined in the offshore GOA and is not significantly influenced by growth in early-marine or of the data produced regression models that were not parsimo- strait habitats (Briscoe 2004). nious, indicating strongly contradictory relationships between Findings from these studies led us to extend the analysis even- and odd-year series. We observed that coho weight was from a single growth-related response variable (adult size) to positively correlated with the PDO Index in even years but explore relationships with survival-related response variables not in odd years, while coho weight was negatively correlated including marine survival, sex ratio, and the per capita re- with the catch of pink salmon in odd years but not in even productive capacity of a coho salmon population. We tested years (Shaul et al. 2011). The need for a parsimonious ex- the set of predictive variables that best explained adult coho planation for coho weight, based on a consistent relationship weight with growth and survival-related response variables with potential causal factors, led us to use multiple regression specific to the Berners River in Southeast Alaska. We also techniques to explore the hypothesis that pink salmon abun- examined relationships between growth-related and surviv- dance and atmospheric forcing are both influential, but that al-related variables for evidence of growth-related late-ma- the effects on coho growth are lagged. A lagged competitive rine mortality to further test the hypothesis that there exists relationship would be consistent with research findings point- an important growth-related late-marine period for survival. ing to an ontogenetic shift in the diet of maturing pink salmon from zooplankton to squid at a weight of about 1,000 g (Aydin 2000; Davis 2003), a size not achieved until late June, on av- METHODS erage, after coho salmon have already fed for several months on the same prey cohort and have achieved nearly two-thirds In the first stage of the analysis, multiple regression of their final weight (Ishida et al. 1998; Fig. 3). The effect models were constructed to explore relationships between of this late transition in diet by pink salmon likely limits the adult coho salmon weight (1970–2014) and potential ex- effect on coho salmon growth of direct competition for the planatory variables at various lags to test our hypothesis that current-year squid cohort by the current-year pink salmon co- variation in coho weight can be explained by lagged effects hort, suggesting that the observed intensifying biennial cycle of climatic variation and top-down control on squid prey in coho size may reflect changes in prey populations that have populations. Software used to run the analysis was R (ver- developed over sequential generations. sion 3.2.3) (R Core Team 2015). We then extended the analysis to examine evidence for growth-related late-marine mortality through effects on ma- Data Sources rine survival, sex ratio, and per capita reproductive poten- tial. Several studies have pointed to an early marine critical Response variables used in the analysis were obtained period for survival of coho salmon within the first weeks or from two sources, (a) commercial catch data showing the av- 331
NPAFC Bulletin No. 6 Shaul and Geiger Table 1. Description of explanatory and response variables and data sources. Explanatory variables Description/Source Pacific Decadal April-March average of monthly PDO index values ending in the year of maturity for coho salmon. The monthly Oscillation (PDO) index data series is maintained by Nate Mantua (University of Washington): http://research.jisao.washington.edu/pdo/ PDO.latest Commercial catch Commercial catch by species in North America (excluding NPAFC area W-AK, the Aleutian Islands and Bering of pink and sockeye Sea) in metric tons; 1964–2011 data are available in Irvine et al. (2012); 2012–2014 catches for Canada, salmon Washington and Oregon were downloaded as a statistical data file from the NPAFC: www.npafc.org/new/ science_statistics.html Alaska catches in 2012–2014 (excluding the Aleutian Islands and Bering Sea) were provided by Kurt Iverson, Alaska Department of Fish and Game, Commercial Fisheries Division, Juneau. Response variables Description/Source Coho weight The weekly total weight of head-on, gutted coho salmon landed by the Southeast Alaska troll fishery divided by the number of fish reported in the landings. Weekly average weights were averaged over a period of 11 statistical weeks (weeks 28-38) from early July to mid-September. Data were accessed from the catch data base using the Alaska Department of Fish and Game’s ALEX program and are reported by Shaul et al. (in press). Adult length Average mid-eye to fork length of male and female age-.1 coho salmon spawners in the Berners River estimated prior to the gillnet fishery. (Shaul et al. in press). Sex ratio Number of females-per-male estimated prior to the gillnet fishery. (Shaul et al. in press). Marine survival Total return (harvest plus escapement) of age-.1 coho coho salmon returning to the Berners River divided by the number of smolts emigrating in the prior year. (Shaul et al. in press). Egg biomass per female Average egg biomass of female Berners River coho salmon (prior to the gillnet fishery) based on an estimated relationship between female length and egg biomass reported by Fleming and Gross (1990) and Shaul et al. (in press). Per Capita Egg Estimated egg biomass per female Berners River coho salmon multiplied by the estimated proportion of females Biomass (PCEB) in the population prior to the gillnet fishery. (Shaul et al. in press). erage weight of troll-caught coho salmon in Southeast Alas- passages, and a gillnet fishery conducted near the river. In ka during 1970–2014 and (b) growth and survival-related order to account for size selection in the latter fishery, we re- variables specific to the Berners River population for adult constructed the pre-fishery length distribution and computed returns in 1990–2014 (Table 1; Shaul et al. in press). average length (following Kendall and Quinn 2012), using Coho weight was calculated by dividing the weight of length measurements from an average of 339 coded-wire head-on, gutted coho salmon landed by the Southeast Alas- tagged Berners River fish sampled annually from the catch. ka troll fishery by the associated number of fish reported on Sex was not determined for the catch, so estimation of the sales slips. There is a seasonal trend of increasing average effect of the harvest on the sex ratio required an assumption weight, as well as substantial inter-annual variation in the that fish of the same length were equally vulnerable to the temporal distribution of the troll catch (Shaul et al. 2011). fishery, independent of sex. Therefore, average weight was calculated weekly and aver- Per capita reproductive potential was assumed to be aged across 11 statistical weeks (weeks 28–38), spanning a proportionate to the per capita egg biomass (PCEB). We period from early July through mid-September, in order to used an average relationship between egg biomass (EB) and obtain a temporally stable measure of average coho salmon female length from two British Columbia coastal streams, weight in coastal waters. Mamquam River and Tenderfoot Creek (Fleming and Gross Marine survival and the size and sex composition of 1990). Letting MEF denote the mid-eye to fork length age-.1 returning adults were estimated annually for 1990– (mm), the following is the conversion relationship applied 2014 adult coho salmon returns to the Berners River, located to females in the Berners River: 65 km north of Juneau, Alaska (Shaul et al. in press). A tar- get sample of 600 spawners was captured from upper river = 2.33 × 10−7 [ ]3.39 . pools using a 13.7-m beach seine and sampled for age, sex, and mid-eye to fork (MEF) length. Returns to the Bern- Estimates of egg biomass for individual females were ers River are comprised almost entirely of age-.1 adults that averaged and multiplied by the proportion of females in have spent one year at sea, with age-.0 jacks being rare. Ma- the adult population to estimate PCEB, which was then rine survival was estimated by dividing the total age-.1 adult converted to a PCEB index by dividing the annual value by return (combined catch and spawning escapement estimates) the average for all 25 years. by the estimated smolt migration in the prior year. Explanatory variables included the commercial catch Returning fish are exploited intensively by two major of pink and sockeye salmon (in metric tons) as a measure fisheries, including a troll fishery in outer coastal waters and of the biomass of maturing salmon (Table 1) and, by infer- 332
Climate and competition effects on coho salmon NPAFC Bulletin No. 6 ence, the potential for each species to influence availability one lag, we considered averages across lags to develop new of squid prey for coho salmon. Biomass of the catch was se- explanatory variables. We tested models that included sock- lected over numerical abundance as an explanatory variable eye salmon catch as a separate variable from pink salmon because biomass includes elements of both abundance and catch, as well as the pooled catch of both species under the size. Evidence of a strong positive relationship between the assumption of an equal effect (per unit of weight) on the prey individual size of pink and sockeye salmon and the amount species of interest. Each predictive time series was standard- of squid in their diet (Aydin 2000; Davis 2003) suggests that ized (the mean of the values actually used in the regression total biomass is a more accurate measure of the potential relationship was subtracted and the result was divided by the for both species to influence squid prey populations of im- sample standard deviation). Model residuals were tested for portance to coho salmon. Salmon biomass variables tested autocorrelation using a Durbin-Watson test and by examining included separate values for pink and sockeye salmon, as the sample autocorrelations. Models were ranked in order well as the combined biomass of both species. We used the with the change in Akiaike Information Criterion differences combined commercial catch in North America, excluding (ΔAIC; Burnham and Anderson 1992). Models with ΔAIC ≤ fishing areas in the Aleutian Islands and Bering Sea, with 2 were considered to have equivalent support. the objective of indexing the biomass of pink and sockeye We tested the combination of predictive variables for salmon maturing primarily within the GOA. coho weight with the lowest ΔAIC score in models explain- North Pacific climate was represented by a single vari- ing adult length, sex ratio, PCEB index, and marine survival able, the 12-month (April–March) average monthly PDO for the Berners River population. Multiple regression mod- index ending in the coho salmon catch year. This period els were developed for length of adults of each sex and the was targeted to encompass the period of hatching and devel- mean-average of both sexes prior to exposure to the gillnet opment for B. anonychus, based on the occurrence of new fishery. Additional models were developed for marine sur- paralarvae in the northern GOA beginning in April and as- vival, ratio of females-to-males, and the PCEB index. Sin- suming a 2-year lifespan (Jorgensen 2011). gle-variable regression models were also used to explore re- lationships between the catch of pink salmon and variables Models representing adult length, sex ratio, PCEB index, and ma- rine survival for the Berners River population, as well as Multiple regression analysis was used to explore rela- relationships among growth and survival-related variables. tionships between coho salmon weight, as the response vari- Additionally, these variables were also differenced so as to able, and the PDO index and pink and sockeye salmon catch- show the relationship between pink salmon biomass and the es at various lags ranging from 0 to 6 years from the catch year-to-year change in adult length, sex ratio, PCEB index, year for adult coho salmon. Each predictive series was test- and marine survival for the Berners River population. Re- ed for obvious autocorrelation structure using conventional lationships among response variables were plotted and ex- time-series analysis tools, including calculating the sample amined separately for the second half of the series (2002– autocorrelations and partial autocorrelations out to at least 12 2014), which occurred after a shift to a cooler North Pacific lags. Cross-correlation values were generated between the climate (Peterson and Schwing 2003). coho weight series and each of the other variables to see at We rearranged the top-ranked predictive model for which lags the variables might be most useful for predicting coho weight to examine the effects of salmon biomass dependent variables. In cases with correlation at more than separately from climate, and to estimate a climate-based Table 2. Model selection statistics for analyses of hypotheses for average weight of troll-caught coho salmon, 1970–2014. Terms in the hypotheses are the commercial catch of pink salmon or pink and sockeye salmon combined (in millions of fish) and the April–March average monthly Pacific Decadal Oscillation (PDO) index ending in the coho salmon return year. The independent variables are lagged from 0 to 4 years (denoted 0, -2 or -4). Models are ranked by the Akiaike Information Criterion differences (ΔAIC). Models with ΔAIC ≤ 4 are listed, with the best model shown at the top. Coefficient weights Adjusted Hypothesis R2 ΔAIC Salmon PDO R2 Pink (average -2, -4) + PDO (average 0, -2, -4) 0.508 0.492 0.646 0.629 0.00 Pink & Sockeye (average -2, -4) + PDO (average 0, -2, -4) 0.483 0.517 0.644 0.627 0.29 Pink (-2) + Pink (-4) + PDO (average 0, -2, -4) 0.522a 0.478 0.651 0.625 1.72 Pink & Sockeye (-2) + Pink & Sockeye (-4) + PDO (average 0, -2, -4) 0.497a 0.503 0.648 0.623 2.00 Pink & Sockeye (average 0, -2, -4) + PDO (average 0, -2, -4) 0.504 0.496 0.625 0.608 2.53 Pink (-2) + PDO (average 0, -2, -4) 0.494 0.506 0.618 0.600 3.38 Pink & Sockeye (-2) + PDO (average 0, -2, -4) 0.474 0.526 0.615 0.597 3.75 Pink (average 0, -2, -4) + PDO (average 0, -2, -4) 0.504 0.496 0.615 0.597 3.76 a Coefficient weights at specific lags are: Pink (-2): 0.329, Pink (-4): 0.193, Pink & Sockeye (-2): 0.312, Pink & Sockeye (-4): 0.185. 333
NPAFC Bulletin No. 6 Shaul and Geiger carrying capacity for the GOA to produce pink salmon for tial. The third highest ranked model included pink salmon harvest, given an objective of maintaining a constant av- biomass at separate lags of 2 and 4 years, with the lag 2 erage coho salmon size. The regression model describing coefficient weight (0.329) comprising 63% of the total co- coho weight (W) as a function of the pink salmon catch efficient weight assigned to salmon (0.522) while the lag biomass (Pink), the PDO index (PDO), and a random (un- 4 coefficient weight (0.193) accounted for 37%. The top correlated) normally distributed error (ε), where b1 and b2 ranked model (hereafter this predictor set will be referred are respective variable coefficients and c is a constant, is to as the Pink-PDO predictors) included the pink salmon shown as follows: catch biomass averaged over the two prior cycles (lag 2 and 4 years; Fig. 5). No significant autocorrelation was detect- = ( 1 ) + ( 2 ) + + . ed in the residuals for this model at lags of 1–15 years and the Durbin-Watson statistic was not significant (p = 0.474). By ignoring the error and fixing coho weight (W) at a Partial residual plots indicate a strong negative relationship constant value (in this case the 45-year average of 3.09 kg), with pink salmon biomass (Fig. 5C) and a strong positive we can rearrange the model to estimate the capacity (K) of relationship with the PDO index (Fig. 5D), with 1995 and the GOA to produce pink salmon for harvest while achieving 1999 appearing as principal outliers. the coho weight target under observed climatic conditions (PDO index) associated with the same coho return year: Climate-Based Capacity 3.09 − ( 2 ) − PDO-based estimates of the climate-based capacity ̂ = . 1 of the GOA to produce pink salmon biomass for harvest while maintaining a constant average target coho salmon weight (3.09 kg) are highly variable, ranging over an order RESULTS of magnitude from a low of 24.1 thousand metric tons in 1976 to 245.7 thousand metric tons in 1998 (Fig. 6). The Coho Weight Model relationship between the estimated climate-based capacity (K̑ ) for pink salmon harvest (in metric tons) at the 3.09 kg Positive autocorrelation was detected in the data series at lags of 1, 2, 3, 4, and 6 years for pink salmon biomass, 1–5 years for sockeye salmon biomass, and 1–7 years for combined biomass of the two species, while the PDO index $ 3LQN 6DOPRQ had significant positive autocorrelation only at lag 1. The best models explaining troll coho weight included salmon abundance and climate variables only in the current year and at lags in 2-year increments up to 4 years (Table 2). Diagnostics for the best models were generally acceptable, with no detected autocorrelation in the residuals (diagnos- &RHIILFLHQW tic checks included calculating the autocorrelation in the residuals out 12 lags, plotting the fitted variables against the residuals, examining Q-Q plots, looking for large lever- age in the residuals, and calculating the Durbin-Watson statistics). Models that included salmon biomass or cli- % 3'2 mate variables for the alternate biennial cycle (at lags of 1 or 3 years) ranked poorly, consistent with a 2-year life cycle in B. anonychus (Fig. 4). Models that included sock- eye salmon as a variable separate from pink salmon did not rank high. Among the four top-ranked models considered to have equivalent support (ΔAIC ≤ 2), two that included the combined biomass of pink and sockeye salmon as a single variable ranked slightly below similar models that included only pink salmon. Highest ranked models con-
Climate and competition effects on coho salmon NPAFC Bulletin No. 6 0.5 3.8 B C Residual Coho Weight (kg) 3.6 0.0 3.4 1995 -0.5 3.2 7 0 7 4 7 8 8 2 8 6 1 9 0 1 9 4 1 9 8 2 0 2 2 0 6 1 0 1 4 1 9 1 9 1 9 1 9 1 9 2 0 2 0 3.0 3.8 A Observed 2.8 2.6 3.6 1999 Model 2.4 0 50 100 150 200 250 Coho Weight (kg) 3.4 Pink Salmon (1,000s of metric tons) 3.2 3.8 Coho Weight (kg) 3.0 3.6 1995 3.4 2.8 3.2 3.0 2.6 R2 = 0.646 2.8 Adjusted R2 = 0.629 1999 2.4 2.6 D 2.4 1970 1974 1978 1982 1986 1990 1994 1998 2002 2006 2010 2014 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 Year PDO Fig. 5. Southeast Alaska troll-caught coho salmon average dressed weight compared with modeled weight (A) based on a multiple regression model with two variables: the standardized April–March PDO Index (average for lag 0, 2, and 4 years; 0.492 weighting based on the regression coefficient) and the standardized average commercial catch of pink salmon in North America (excluding the Bering Sea and Aleutian Islands) lagged by 2 and 4 years (0.508 weighting). The model residual is shown (B), as well as partial residual plots for pink salmon (C) and the PDO index (D). target coho weight is shown by the following relationship Climate-based capacity estimates based on the target with the PDO index: coho weight were exceeded only a few times in even years and by modest percentages prior to 2012, when a series of ̂ = 86.928( ) + 128,066, low trailing PDO index values and substantial even-year pink salmon returns were associated with biomasses that ex- where K at a neutral (0) PDO index value is estimated at ceeded capacity estimates by 102% in 2012 and 83% in 2014 128,066 metric tons, an amount that has been consistently (Fig. 6). Coho weight was the lowest on record for an even equaled or exceeded by the lagging pink salmon catch bio- year in 2012, and third lowest in 2014 (Fig. 2). Since the mass variable since 1987. early 1990s, differences between pink salmon biomass and Although not significantly correlated over the full time estimated capacity have been greater in odd years as odd- series (r = 0.264; p = 0.079), pink salmon catch biomass and year biomass transitioned from being consistently below es- estimated climate-based capacity (i.e., scaled PDO index vari- timated capacity during 1971–1991 (by an average of 26%) able) showed strong positive correlation during 1970–1990 (r to consistently above capacity by an increasing margin since = 0.809; p < 0.001), with capacity exceeding catch biomass in 1993 (158% in 2009, 205% in 2011, 364% in 2013; Fig. 6). all but 2 years. However, pink salmon biomass and estimated capacity were essentially uncorrelated in the subsequent pe- Adult Length riod from 1991–2014 (r = -0.148; p = 0.490), as biomass re- mained high while the PDO index trended lower. This change During 1982–2014, Berners River spawners of both sexes was associated with a substantial (43%) increase in variation declined in length by an average of 1.6 mm/year for males and in annual coho weight (Fig. 5A). However, the model fit was 1.1 mm/year for females (Fig. 7). Variation in length among consistent between the periods (Fig. 5B), with no meaningful spawners returning in the same year increased for both sexes. change in the average residual between 1970–1990 (-0.014) Males showed substantially greater intra-annual variation in and 1991–2014 (0.012), or in coefficients of variation (CV) in length among individual spawners (average CV = 0.109) com- the residuals (0.164 and 0.188, respectively). pared with females (average CV = 0.059) as well as greater in- 335
NPAFC Bulletin No. 6 Shaul and Geiger 260 Climate-Based Pink Salmon 240 Capacity 220 200 Metric Tons (1,000s) 180 160 140 120 100 80 60 Catch Biomass of Pink Salmon 40 Capacity at Constant Target Coho Weight 20 Capacity at Target Coho Weight at Neutral PDO 0 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Year Fig. 6. Average Gulf of Alaska pink salmon catch in the preceding two cycles (lag 2, 4) compared with the estimated catch at a constant target coho salmon weight of 3.09 kg (45-year average) at both the trailing 3-cycle average PDO index (lag 0, 2, 4) and at a constant neutral PDO index. The PDO variable is converted to an estimate of the climate-based capacity of the Gulf of Alaska to produce pink salmon for harvest while also achieving a target coho salmon weight. ter-annual variation in average length (CV = 0.044) compared (Fig. 8A), but recent relationships between length and sex with females (CV = 0.028). During 1998–2010, average ratio and between marine survival and PCEB index exhibit length of both sexes became increasingly cyclical, declining greater slope. Variation in the length-survival relationship in odd years while remaining relatively stable in even years decreased at smaller adult sizes, suggesting a more limited until 2012, when even-year length decreased sharply. range of survivals for cohorts with slower growth, as a po- The linear selection differential (LSD), the difference tential consequence of size-selective mortality (Fig. 8A). in average length before and after the gillnet fishery, aver- During 1990–2014, there were important differences aged -12.3 mm for males and -3.7 mm for females during between even and odd years in the length of age-.1 adults, 1990–2014 (Shaul et al. in press). On average, the estimated the female-to-male ratio, and the PCEB index (Fig. 9). Av- effect of the gillnet fishery on the ratio of females-to-males erage marine survival estimates in odd years (14.6%) were was not meaningful, with the average ratio before and after not significantly different from even years (17.9%; p = the fishery decreasing from 0.80 to 0.75. 0.157). However, the relative survival of females (female- to-male ratio) was lower in odd years (p = 0.012) with a Relationships Between Population Variables pre-gillnet female-to-male ratio of 0.71 compared with 0.88 in even years (assuming a 1:1 sex ratio in smolts; Spidle et There was a moderate correlation between marine sur- al. 1998). The PCEB index was also significantly different, vival and adult length (Spearman’s rho = 0.669, p < 0.001; averaging 18% lower in odd years prior to the gillnet fishery Fig. 8A). The correlation between adult length and the ra- (p = 0.002) and 23% lower in the spawning escapement (p tio of females-to-males was lower, with greater variability < 0.001). in the female-to-male ratio at larger adult length (Fig. 8B). The correlation between marine survival and the female-to- Pink Salmon and PDO Predictors male ratio was considerably lower, and did not reach statis- tical significance (Fig. 8C). The PCEB index, which has as The Pink-PDO predictors consistently explained at least factors both female length and the proportion of the adult a moderate amount of the variation in average size of return- population comprised of females, had a small to moderate ing coho salmon of both sexes in 1990–2014 (Table 3; Fig. correlation with marine survival (Fig. 8D). 9A). Results were consistent with troll weight (1970–2014) The regression slope for the 2002–2014 length-survival in indicating an approximately equal split between pink relationship did not differ from the slope for the entire series salmon biomass and climate (PDO) as factors influencing 336
Climate and competition effects on coho salmon NPAFC Bulletin No. 6
$ $YHUDJH/HQJWK % &RHIILFLHQWRI9DULDWLRQ
/HQJWK PP
&9
0DOHV
)HPDOHV
NPAFC Bulletin No. 6 Shaul and Geiger 35 A 1.3 1.2 B Female-to-Male Ratio 30 Marine Survival (%) 1.1 25 1.0 20 0.9 15 0.8 0.7 10 0.6 5 0.5 1990-2001 1990–2001 2002-2014 2002–2014 1990-2001 1990–2001 2002-2014 2002–2014 0 0.4 575 600 625 650 675 575 600 625 650 675 Length (mm) Length (mm) 1.3 1.2 C 1.4 D Female-to-Male Ratio 1.1 PCEB Index 1.0 1.2 0.9 0.8 1.0 0.7 0.6 0.5 0.8 0.4 1990-2001 1990–2001 2002-2014 2002–2014 1990-2001 1990–2001 2002-2014 2002–2014 0.3 0.6 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 Marine Survival (%) Marine Survival (%) Fig. 8. Regression relationships for age-.1 Berners River coho salmon: marine survival vs. length (A), sex ratio vs. length (B), sex ratio vs. marine survival (C), and PCEB index vs. marine survival (D). Earlier years (1990-2001) are shown with gray dots and later years (2002–2014) with black dots. Linear relationships for the entire data series are shown with solid lines while relationships for 2002-2014 only are indicated with dashed lines. Length, sex ratio and PCEB index values are estimated prior to the gillnet fishery and length values are the mean-average for both sexes. Spearman’s rho correlations are significant (p ≤ 0.05) for all relationships except for sex ratio vs. survival for the full data series (1990– 2014). Note: sex ratio (C) and the PCEB index (D) include elements of survival and are, therefore, not entirely independent from marine survival. ble, with approximately equal weighting indicated for top- One obvious criticism of our approach is that both down control (0.508) and climate (0.492) variables targeted the predictive and response variables contain autocorrela- at squid recruitment and survival. Our results are in strong tion. The important effect of this is to potentially produce agreement with Jorgensen’s (2011) hypothesized 2-year misleading error rates in statistical hypothesis tests of zero lifespan for B. anonychus, as well as our hypothesis that correlation (e.g., Pyper and Peterman 1998). However, our coho salmon size reflects a lagged response by reproductive- intent was never to simply test the hypothesis that there was ly isolated even- and odd-year populations of B. anonychus zero correlation between any two variables. Rather, we were to variable intensity of top-down control by pink salmon. looking for consistent relationships between coho salmon The most likely explanation for the lagged response (Fig. 4) size and environmental and competition metrics—consistent is a related delay in predation on maturing squid by maturing over a period of improving environment (from 1970 to the pink salmon that limits the effect on coho salmon growth of early 1990s) and a period of declining environment (mid- direct competition for the current prey cohort. Pink salmon 1990s to the present, Fig. 6). We did not attempt to adjust appear to influence coho salmon growth primarily through error rates or p-values, but rather we were guided by the no- predation on the parents and grandparents of the current tion that we were especially skeptical of any hypothesis tests squid cohort, with the parent generation being most import- that were not highly significant using conventional p-value ant (accounting for 63% of the combined pink salmon coef- calculations. In the end, we found essentially the same Pink- ficients). PDO predictor signal in different measures of coho size, and 338
Climate and competition effects on coho salmon NPAFC Bulletin No. 6 coho size consistently trended upwards with increases the R2 $ $JH $GXOW /HQJWK PDO metric and downwards with increases in the pink salm- /HQJWK PP on metric, in a way that was consistent along both even- and odd-year lines. The two predominant outlying years when coho salmon 2EVHUYHG 0RGHO weighed substantially more (1995) and less (1999) than in- (YHQ\HDU$YHUDJH 2GG
NPAFC Bulletin No. 6 Shaul and Geiger A (Sex Ratio) C (PCEB Index) E (Marine Survival) 1.3 1.3 35 1.2 1.2 30 Females Per Male 1.1 PCEB Index 1.1 25 Survival (%) 1.0 0.9 1.0 20 0.8 0.9 15 0.7 0.8 10 0.6 0.7 5 r = -0.576 0.5 r = -0.329 r = -0.484 0.4 0.6 0 120 160 200 240 120 160 200 240 120 160 200 240 Salmon (1,000s of tons) Salmon (1,000s of tons) Salmon (1,000s of tons) B (Change in Sex Ratio) D (Change in PCEB F (Change in Survival) 0.6 0.6 15 Change in Index Index) 0.4 0.4 10 Change in Ratio Change (%) 0.2 0.2 5 0.0 0.0 0 -0.2 -0.2 -5 -0.4 -0.4 -10 r = -0.733 r = -0.767 r = -0.613 -0.6 -0.6 -15 120 160 200 240 120 160 200 240 120 160 200 240 Salmon (1,000s of tons) Salmon (1,000s of tons) Salmon (1,000s of tons) Fig. 10. Relationship between the average pink salmon harvest (lag 2 and 4 years) and the sex ratio (females per male; A), the year-over- year change in sex ratio (B), the PCEB index (C), the change in the PCEB index (D), the marine survival rate (E), and the change in marine survival (F) for coho salmon returning to the Berners River, 1990–2014. Sex ratio and PCEB index values are prior to exposure to the drift gillnet fishery. scale growth (Briscoe 2004) in indicating that adult size is by the Pink-PDO predictors that explain much of the vari- influenced primarily by conditions encountered in offshore ation in adult size. The model containing these variables waters of the GOA. together accounted for over a third of variation in marine The moderately strong positive correlation between survival (R2 = 0.378) but the PDO coefficient was not signif- marine survival and size of Berners River adults is, there- icant. Although the pink salmon predictor alone was signifi- fore, consistent with the hypothesis that overall survival in cant (r = -0.576; Fig. 10E), model diagnostics were poor with the ocean is related to late-marine growth. The evident de- the variance of the residuals decreasing with increases in the crease in variation in survival at smaller adult sizes (Fig. 8A) predictor and with significant autocorrelation in the residuals. suggests that slower late-marine growth may reduce both In contrast, a direct relationship with adult length explained average survival and the potential range of survival rates. somewhat more of the variation in marine survival and the This suggests that as the rate of growth slows in the offshore model diagnostics were better. This may mean that varia- environment, growth-related late-marine mortality may be- tion related to late-ocean growth accounted for some of the come a proportionately more important influence on marine variation in marine survival for the Berners River population, survival compared with other factors. but that the link to the PDO and pink salmon is less direct. Our model indicates about half of the nearly two-thirds Potential countervailing effects (perhaps less growth-related) of variation in adult size explained by the coho weight model by the predictive variables on survival should be considered, is attributed to the biomass of pink salmon in the GOA while however, we found no evidence of a positive relationship the other half is attributed to climatic factors related to at- with indicators of abundance of pink salmon (or hatchery mospheric forcing (measured by the PDO index). However, chum salmon) in near-shore environments. when the same Pink-PDO predictors from models explaining The pink salmon catch (average for lags of 2 and 4 adult size were applied to survival-related response variables, years) included in the pink-PDO predictors explained much only the pink salmon biomass variable showed a consistent of the year-to-year change in marine survival, sex ratio and statistically significant influence. For example, marine sur- PCEB index (with acceptable model diagnostics) suggesting vival for the Berners River population was poorly explained that while trends in marine survival may be influenced by 340
Climate and competition effects on coho salmon NPAFC Bulletin No. 6 other factors, the biomass of pink salmon has an important $ )HPDOHWR0DOH5DWLR R2 effect on year-to-year variation in survival of coho salmon. &KDQJH We infer that the probable underlying mechanism is control of squid prey populations by pink salmon. More recently (2002–2014), variables associated with 2EVHUYHG 0RGHO growth were more strongly correlated with marine survival (YHQ
NPAFC Bulletin No. 6 Shaul and Geiger 3.5 630 between marine survival and size-at-maturity for Berners 3.4 620 River coho salmon. 1995 The timing and mechanisms underlying late-marine 3.3 610 mortality remain unclear. While maturing squid (B. anony- 3.2 600 chus) have been found to comprise the majority of the sum- Length (mm) Weight (kg) 3.1 590 mer diet of coho salmon above a weight of 500 g (Davis 3.0 580 2003), a size that is reached on average in January (Ishida et al. 1998), B. anonychus may also be important in the diet 2.9 570 of coho salmon during winter months when growing squid 2.8 560 are also smaller (Aydin 2000). If so, variation in growth-re- 2.7 Observed Troll Weight 550 lated mortality linked to squid abundance may begin during 2.6 Modeled Troll Weight 540 winter from a physiologically based process (Beamish and 2.5 530 Mahnken 2001). Unfortunately, this hypothesis is difficult Offshore Length 1999 to assess because of a scarcity of information on the winter 2.4 520 1994 1995 1996 1997 1998 1999 2000 diet and condition of coho salmon in offshore waters. Predation appears to be the most likely cause of mor- Year tality of maturing fish during summer. While females may Fig. 12. Observed and modeled average weight of troll-caught take greater risks with predators when food is scarce be- coho salmon (with 1995 and 1999 outliers indicated; see Fig. 5) cause of their greater energy and growth requirement for compared with average fork length of fish sampled at offshore stations along longitude 145°W (Kaeriyama et al. 2004). successful reproduction (Holtby and Healey 1990), a sub- stantial proportion of males may be motivated by similar pressures. The large amount of variation in size of age-.1 pink salmon. Ruggerone et al. (2003, 2005) and Ruggerone males appears to stem from disruptive selection associated and Connors (2015) presented evidence indicating that with the option of two viable breeding strategies: stealth growth and survival of sockeye salmon returning to Bris- (satellite) or dominance (alpha; Healey and Prince 1998). tol Bay and the Fraser River, respectively, was reduced by The largest (as well as smallest) individuals returning to the a competitive interaction with pink salmon occurring pri- Berners River are invariably males, suggesting that larger marily in the second year at sea. The reduction in apparent males have also expressed a willingness to trade survival growth in odd years for Bristol Bay sockeye salmon occurred for growth in order to be competitive as dominant spawners, in summer after highly abundant Russian pink salmon popu- even as a substantial proportion of males may pursue an lations had migrated to coastal areas, an effect that may have opposite strategy in years of poor growth when a reduced been reinforced by a biennial cycle in prey, including squid female-to-male ratio likely enhances the advantage to mid- (Ruggerone et al. 2005; Ruggerone and Connors 2015). dle-sized males of trading growth for survival, thereby ac- Wide-spread declines in abundance of Chinook salm- cepting a stealth role over dominance in a more competitive on populations have occurred throughout Alaska since 2007 breeding environment. (ADF&G 2013) concurrent with consistent over-prediction Specific mechanisms behind risk-taking as a cause of by sibling-based forecast models for stocks contributing late-marine mortality are poorly understood but may include to Southeast Alaska fisheries (CTC 2014). Broad declines some combination of increased metabolic cost relative to since the early 1980s have been documented in size-at-age reward and increased exposure to salmon predators while and age-at-maturity of Alaska Chinook salmon populations undertaking searching movements or while pursuing prey in (Kendall and Quinn 2011; Lewis et al. 2015). The steep- the vicinity of “patches” of food that may concentrate biota est declines in size-at-age have occurred in older fish, pri- at multiple trophic levels (Benoit-Bird and Au 2003). Spa- marily those that have spent four years at sea, while age-.2 tial variation in food and risk factors may occur across dif- fish have shown little change. A combination of decreasing ferent geographic scales, from intensive patches of mesozo- size-at-age and decreasing age-at-maturity is unexpected, as oplankton (Russell et al. 1992) to the scale of oceanographic Chinook salmon have been shown to delay maturity when domains. Salmon dietary studies indicate that B. anonychus growth is poor (Healy 1991; Wells et al. 2007). typically appears in higher density in the Subarctic Current As in coho salmon, the decrease in apparent growth and compared with other North Pacific domains (Davis 2003; survival of Chinook salmon is potentially related to a decline Kaeriyama et al. 2004), while on a smaller scale, the spe- in gonatid squids, which are typically the dominant prey of cies has been found concentrated above seamounts (Nesis older Chinook salmon in offshore waters of the northeast 1997). Depending upon their persistence, such aggregations and north-central Pacific and the Bering Sea (Davis 2003; may attract not only higher trophic level salmon species, Kaeriyama et al. 2004; Davis et al. 2009). Evidence point- but species such as salmon sharks, which are abundant and ing to an increase in late-ocean mortality as a factor in de- effective predators on maturing salmon (Nagasawa 1998) clines in Chinook salmon abundance since the mid-2000s and also feed extensively on B. anonychus and other squids is consistent with an increase since 2002 in the correlation (Kubodera et al. 2007). 342
Climate and competition effects on coho salmon NPAFC Bulletin No. 6 Climatic Effects shore subarctic waters. In contrast with our results, size vari- ation of coho salmon stocks south of Alaska has been shown The occurrence of biennial lags in both of the Pink-PDO to be negatively correlated with warm conditions (positive predictors (Fig. 4) leads us to infer that the connection be- PDO; Wells et al. 2006), while recruitment of natural coho tween the PDO and coho weight likely occurs through a cli- salmon from Oregon coastal rivers showed a strong negative matic link to recruitment of squid. The positive association correlation with the spring/summer PDO averaged over a pe- between coho weight and the PDO index across multi-gen- riod of four years prior to the return year (Rupp et al. 2012). erational lags, with no evidence of influence during off-cycle years, suggests that B. anonychus survival is closely cou- Interactions with Pink Salmon pled with atmospheric forcing in the North Pacific. Potential mechanisms include improved early survival in response to An important feature of the relationship between pink more abundant food associated with a shallower mixed lay- salmon biomass and estimates of climate-based capacity er (Polovina et al. 1995) and improved transport of larvae (i.e., scaled PDO index variable) is their transition from be- by currents to locations favorable for survival (Bailey and ing strongly correlated with each other during 1970–1990 to Picquelle 2002) during conditions associated with a strong being uncorrelated afterward (Fig. 6). The marked change Aleutian Low and high PDO index values. The distribution in the relationship between atmospheric forcing and pink of squid within the Alaska Gyre, as well as their abundance, salmon returns in the northeast Pacific was associated with may be linked to physical oceanographic variables (Aydin a substantial increase in variation in annual coho weight. et al. 2000). We infer from these results that a decrease in synchrony However, other plausible mechanisms may contribute between variables representing bottom-up (positive) and to the observed positive relationship between adult coho top-down (negative) influences has increased vulnerability weight and the PDO. Exceptionally warm climatic condi- of epipelagic squid populations to steep declines during pe- tions in the northeast Pacific in 1997 and 2015 were associ- riods when both factors are unfavorable for survival. Our ated with peaks in average size of juvenile coho salmon sam- model for estimating the climate-based capacity of the GOA pled during late-July in trawl surveys in northern Southeast ecosystem to produce pink salmon for harvest while main- Alaska (J. Orsi, joe.orsi@noaa.gov, pers. comm.), suggest- taining coho salmon at a historical average weight provides ing that warm conditions associated with high PDO index a potential template for evaluating some of the ecosystem values are favorable for early-marine growth of coho salmon trade-offs associated with ocean ranching. prior to when they move offshore and begin feeding on B. Other investigators have found evidence of control of anonychus. In addition, results of bioenergetics simulation squid populations by pink salmon, based on opposing bien- indicate that optimal temperatures for growth are positively nial cycles in the western North Pacific and Bering Sea (Ito related to daily rations (Beauchamp 2009), indicating that 1964; Davis 2003). Ogura et al. (1991) observed an even- warmer temperatures associated with high PDO index val- year dominant pattern in length of coho salmon in the west- ues may reinforce the effect of an increase in prey availabil- ern North Pacific that developed during the second summer ity by also increasing the growth response in coho salmon. at sea when diet overlap with pink salmon increases. Our Aydin (2000) estimated that a systemic 10% increase in sea results indicate that a similar interaction exists and has been surface temperature in the vicinity of the squid-rich Subarc- intensifying in the northeast Pacific. tic Current would favor squid-feeders like coho salmon, as Wild pink salmon populations appear to have benefited they currently find enough food to benefit from increased from recent climatic patterns and effective fishery manage- metabolic activity associated with warmer water. ment practices. These salmon have remained at high abun- Although specific mechanisms behind the inferred con- dance, particularly in odd years, despite a recent turn in the nection between the PDO and recruitment of B. anonychus North Pacific climate cycle to cold conditions that have his- await further study, it seems likely that other subarctic ceph- torically been associated with poor returns (Beamish and alopods with similar life histories may be similarly influ- Bouillon 1993). Interest in further increasing utilization enced by climate, a factor that should be considered when of offshore salmon forage through ocean ranching of pink investigating causes of variation in growth of other higher salmon (Stopha 2013) underscores the importance of under- trophic level species known to consume cephalopods. For standing the trade-offs at higher trophic levels. Chum salm- example, Wells et al. (2008) found a direct positive relation- on, which have the most distinctive diet among species of ship between apparent growth of Chinook salmon from the Pacific salmon (Welch and Parsons 1993) and consume few Taku River in Southeast Alaska and the Aleutian low pres- maturing squid (Kaeriyama et al. 2004), offer an alternative sure index (closely related to the PDO index used in this to pink salmon for aquaculture that may substantially reduce study) during their 3rd and 4th ocean years, when Chinook the negative effects on higher trophic level species indicated salmon are known to feed heavily upon squid (Davis 2003; by this study. Davis et al. 2009). We hope that our findings help clarify which species Our findings do not appear applicable to more southern (pink salmon or B. anonychus) holds the commanding po- coho salmon populations that do not feed extensively in off- sition in the offshore trophic triangle (Fig. 1). Aydin (2000) 343
NPAFC Bulletin No. 6 Shaul and Geiger A (Local Gulf of Alaska Range) B (Inter-regional Ocean Range) 2.5 2.5 2.0 2.0 Standardized Weight 1.5 1.5 Standardized Length 1.0 1.0 0.5 0.5 0.0 0.0 -0.5 -0.5 -1.0 -1.0 -1.5 -1.5 -2.0 -2.0 -2.5 -2.5 1982 1986 1990 1994 1998 2002 2006 2010 2014 1982 1986 1990 1994 1998 2002 2006 2010 2014 Year Year Pink Salmon Weight Age-1.2 Sockeye Length (5 stocks) Coho Salmon Weight Age-.4 July Troll Chinook Length Fig. 13. Standardized size of salmon in harvests and escapements in Southeast Alaska including (A) weight of commercially-caught pink salmon and troll-caught coho salmon, and (B) length of age-1.2 sockeye salmon spawners (male and female average for Chilkoot River, Situk River, Ford Arm Creek, McDonald Lake, and Hugh Smith Lake) and troll-caught age-.4 Chinook salmon (mean-average for ages 0.4 and 1.4). All slopes are significant (p ≤ 0.05). has suggested that the trophic position and presumed high have decreased, suggesting that increased top-down control productivity of B. anonychus may give it a controlling posi- by salmon (combined with recent unfavorable climatic con- tion in the ecosystem. However, our findings suggest that B. ditions for squid) may be reducing the mean trophic level of anonychus is less productive and more vulnerable to inten- prey in the forage base in a way similar to the phenomenon sive predation by salmon than had been presumed. We infer of “fishing down marine food webs” (Pauly et al. 1998). A from our results that the pink salmon is a keystone predator shortened food chain, with squid reduced as an intermedi- that exerts top-down control (over squid) and thereby directs ate trophic component, may have instead increased energy energy flow in the ecosystem. transfer efficiency between primary production and salmon. Aydin (2000) produced from his extensive investiga- Thus, increased top-down pressure by pink salmon on mi- tion of trophic dynamics and bioenergetics relationships a cro-nektonic squid occupying an intermediate trophic level conceptual model of salmon carrying capacity in the GOA may actually increase the capacity of the GOA to produce that predicts that adding more small salmon to the ecosys- salmon biomass, in direct reversal of the self-defeating hy- tem through ocean ranching may be self-defeating. He pothesis. A key element determining the direction of the hypothesized that introducing increasing numbers of small feed-back response to increasing pink salmon abundance salmon may, through density-dependent effects, reduce their lies in squid populations, which appear substantially less re- early growth rate, thereby delaying their ontongenetic shift silient and more vulnerable to top-down control by salmon to squid prey while placing further pressure on zooplank- than has been assumed. ton. Release of squid from top-down control by their in- An important area for future investigation is to explain tra-guild predator (pink salmon) may lead to an increase in how flexible planktivores have been able to increase in adult squid abundance, placing further demand on zooplankton size in the face of indications of decreased squid abundance, and leading to further decline in salmon growth in a self-de- and in apparent contradiction with bioenergetics results re- feating cycle. ported by Aydin et al. (2005) indicating that the zooplankton Our findings, supported by trends in size-at-age for var- found in the diet of maturing pink salmon do not have the ious salmon species in Southeast Alaska (Fig. 13), are con- caloric density needed to support the apparent growth tra- sistent with the feed-back loop proposed by Aydin (2000) jectory of pink salmon as they approach maturity. Although but suggest that the mechanism has been operating in direct untested, one mechanism that could potentially explain an reverse of his hypothesized self-defeating response. Flexible increase in average size of salmon in the face of a decline planktivores (age-.1 pink salmon and age-.2 sockeye salm- in squid is a temporal advance in the growth curve result- on) have increased in size during 1982–2014 while obligate ing from an increase in abundance or nutritional quality of nektivores (age-.1 coho salmon and age-.4 Chinook salmon) available zooplankton prey beginning earlier in marine life 344
You can also read
